Abstract
The first part of my talk covers a series of studies concerned with the detection aberrant splicing from RNA-seq data [1-2] and their predictions from genomics sequence [3-4]. This work has translational relevance for pinpointing the genetic cause of individuals affected with a genetically undiagnosed rare disorders [5]. The algorithms developed could also help for interpreting mutations found in tumours.
In the second part of my talk, I will present recent results using language models to identify conserved genomic sequences in an alignment-free fashion [6]. We train a masked language model on 3’ regions of ORFs across more than 800 fungal species spanning over 500 million years of evolution. We show that explicitly modeling species is instrumental in capturing conserved yet evolving regulatory elements. We demonstrate the utility of the resulting sequence embedding for range of regulatory genomics prediction tasks.
References: 1. Mertes, Scheller, et al. Detection of aberrant splicing events in RNA-Seq data with FRASER. Nature Communications, 2021; 2. Scheller et al. Improved detection of aberrant splicing using the Intron Jaccard Index. medRxiv 2023; 3. Cheng, et al., MMSplice: Modular modeling improves the predictions of genetic variant effects on splicing, Genome Biology. 2019; 4. Celik, Wagner, et al. Aberrant splicing prediction across human tissues. bioRxiv, 2022 (now in press Nat genet.); 5. Yépez, Gusic, et al. Clinical implementation of RNA sequencing for Mendelian disease diagnostics. Genome Medicine, 2022; 6. Gankin, Karollus, et al. Species-aware DNA language modeling. bioRxiv 2023
Zoom info
Join Zoom Meeting at https://uio.zoom.us/j/65895596084?pwd=b29ubjQ2VEtCNE1vM0NPRkluMGFKdz09.
Meeting ID: 658 9559 6084; Passcode: 505310
Junior talk
Ping-Han Hsieh, PhD candidate at the Centre for Molecular Medicine Norway (NCMM), University of Oslo, will present his work on "Using hierarchical variational autoencoders to incorporate conditional independent priors for paired single-cell multi-omics data integration."